Abstract: Aim at getting the accurate rice planting area and improving the precision of remote sensing monitoring, based on the HJ-1A/B satellite data and crops mask and combined with the measured spectral data, this paper determined the end-members and used linear spectral mixture model (LSMM) for Spectral Unmixing, and obtained the abundance image of mid-season rice and RMS error image in Jianli. After evaluating the mid- season rice planting areas, the result was compared with agricultural statistics and survey data. According to the total accuracy evaluation method, the classification accuracy of rice planting areas reached 93.68%, and according to the sampling accuracy evaluation method, the classification accuracy of rice planting areas reached 83.67%. The results indicated that: the method to determine the end-member selection was feasible, it could be applied to monitor paddy rice information in the plains with HJ-1A/B satellite data and provide information services for the government.